With polls opening in just a few hours there is universal consensus that the Coalition will defeat the Labor government and wield a sizable majority in the new house. My model is predicting a final result of 85 seat to the Coalition and 62 for Labor.

This prediction assumes Labor will not reclaim Melbourne from the Greens, the independents Wilkie and Katter will both hold their seats, and that the Coalition will hold O'Conner, Durack, and Indi from non-coalition conservative challengers.  

The next few paragraphs are basic Australian background and housekeeping related information. If you've read my previous Australian Election modelling diaries you might like to skip straight to the state breakdowns where you'll find the usual tables now include percentage chances for the Coalition to win each seat.

Currently Australia is governed by the Labor party (centre-left, loves unions) with the support of a number of independents (3 conservative, 1 Laborish, 1 idiosyncratic but left-leaning) and the Greens (progressives). The opposition is known as the Coalition, with an uppercase C as their coalition is permanent, and is composed of the Liberals (centre-right, hates unions, loves free markets) and Nationals (conservative agrarian socialists, loves protectionism). The Labor party need the independent and Green support as they do not have enough seats to govern in their own right (and in fact have less seats (71) than the Coalition (72)).

Voting in Australia is compulsory and uses a preferential ballot (this means that in certain seats the way parties direct their supporters to allocate their preferences will be crucial) in single-member seats for the House of Representatives.

Finally before we get to the numbers I'll just mention a few details in regards to the model I am using. The model takes into consideration the prior voting history of the electorate, incumbent strength (where applicable), and public polling. It doesn't account for potential asymmetric swings within states (not enough polling data available). I'm interested to see how much that ends up mattering.

Tables in this diary are colour coded. Shades of red reflect Labor held seats and predictions, shades of blue represent Liberal held seats and predictions (the occasionally different party names and abbreviations are courtesy of local party branches having inconsistent names), independents are grey, Greens are light green (shocker), and Nationals are dark green. The numbers either side of the dash are the two party preferred percentages (Labor-Coalition). The percentage is the odds of the Coalition winning the seat.

New South Wales

New South Wales is the state where Labor is in the biggest trouble and will almost certainly lose the most seats. Seat based polling suggests that the Western Sydney seats are more likely to elect a Coalition representative than the model predicts. Dobell is also widely considered to be certain to switch to the Coalition for seat specific reasons, I'm ok with the model's numbers there though.


The model currently predicts only modest gains for the Coalition in Victoria.

Things are going horribly wrong for the Coalition in Indi where the controversial Liberal incumbent is receiving a strong challenge from a local independent. I suspect the Liberals are going to lose Indi at this point, fortunately for them it is unlikely to effect their chances of forming government.

Both Labor and the Greens are claiming they will win Melbourne. It's a tossup.

The Liberals may win Mallee from the Nationals. A Coalition seat either way.  


Queensland has big question marks over it given recent polling showing significant strength for Palmer's United Party (mining billionaire Clive Palmer has founded his own party and is running as a anti-politician, proudly ignorant of economics). The rise of Katter's Australian Party (an eccentric representative from the outback formed his own socially conservative/protectionist party) also makes things hard to predict.

Given Australia's compulsory preferential voting system it's going to matter a lot how these conservative party voters allocate their preferences. I'm going to guess everything is going to cancel itself out and the model will be pretty close but it is possible Labor's chances in Northern Queensland are understated and in Brisbane are overstated.

Bob Katter will retain his seat.

South Australia

The model predicts a status quo result in South Australia as the most likely result.  

Western Australia

I suspect the model is overstating Labor's chances in Canning and understating them in Perth. The Western Australian Nationals are not technically part of the Coalition and might win either or both of Durack and O'Connor.


Labor seem unexpectedly strong in this iteration of the model. Andrew Wilkie is likely to retain his seat.


I've little faith in the model's ability to call the Northern Territory seats of Solomon and Lingiari accurately, the polling is too weak.


Thanks for following my modeling of Australia's Federal election! I'm working for the Electoral Commission again at the election and won't be around tomorrow. I'll post a full review of the model's accuracy in a few weeks when the election results are finalized.

1:29 PM PT: Election morning polling bad for Labor. Final count now 91-56. I don't have time to update state-by-state, sorry. Have a good election!

Originally posted to Alizarin Indigo on Fri Sep 06, 2013 at 05:53 AM PDT.

Also republished by Community Spotlight.

Your Email has been sent.